valv

Valv provides a secure method for agents to interact with databases, ensuring every user has a protected connection. Key features include: • Programmatic, per-user permission setting • Query validation and scope enforcement • Protection against raw SQL injection • Schema introspection for various databases • Token-aware response limiting Unlike traditional connections where a single access point can expose broad data, Valv establishes unique, policy-driven connections for each agent. This means if a customer service agent queries "where's my order?", the system automatically scopes that query to only the relevant customer data, preventing accidental or malicious access to other users' information. Queries are rebuilt and re-checked against your database schema on the server, ensuring security and data integrity. Valv integrates seamlessly into existing stacks, shipping as an MCP server or a TypeScript library. It supports a wide range of databases including Postgres, MySQL, SQLite, and ClickHouse. Its token-aware design prevents large queries from overwhelming an agent's context or exceeding budget limits. This approach allows agents to perform real analytics—filters, aggregates, time-series, and top-N queries—without the risk associated with direct database access. This tool is ideal for development teams, support operations, and data analysts who need to provide trusted tools with database query capabilities while maintaining stringent security protocols and ensuring data privacy. Give your agents the power to access the data they need, with the confidence that every query is validated and precisely scoped.